[omniORB] ping

[Stefan Seefeld]

Duncan Grisby wrote:
> 
> On Friday 14 January, Stefan Seefeld wrote:
> 
> > I would hope that the ping to a client side object would
> > then simply throw an exception. However, this seems not
> > the case, at least not without any special command line
> > options to the ORB. What should I do to get the desired
> > behavior. Or is there an entirely different approach ?
> 
> Trying to contact an object which no longer exists should result in
> either an OBJECT_NOT_EXIST exception if the object's process is still
> running, or a COMM_FAILURE if the process isn't there any more.
> 
> Normally, you'll receive these exceptions very rapidly, but in the
> case that the host is not reachable by TCP (or the process is
> suspended), the default TCP timeout will block the call for a long
> time. You can make the time shorter with -ORBclientCallTimeOutPeriod,
> or omniORB::callTimeOutPeriod().

yeah, that's precisely what I was expecting. However, it apparently is
not what happens. I'm running the display server and the application on
the same machine. Then I kill the application and expect the server to
notice that with the next ping so it can reclaim the resources, destroy
the application windows etc. However, the ping thread hangs within the
clientContext->ping() method. I played with the timeout parameters, nothing
changes.

Stefan
_______________________________________________________              
              
Stefan Seefeld
Departement de Physique
Universite de Montreal
email: seefelds@magellan.umontreal.ca

_______________________________________________________

      ...ich hab' noch einen Koffer in Berlin...